arXiv subjects

Zhiyin Zhou

Publications and source records attributed to Zhiyin Zhou.

4 recordsLinked to original sources

arXiv2026

RegDivergence-101: An LLM Benchmark for Cross-Jurisdiction Regulatory Contradiction Detection in Life Sciences

Pharmaceutical sponsors developing a drug for both the United States and the European Union must reconcile guidance issued independently by the FDA and the EMA. Where the two agencies require substantively the same thing, a sponsor can file once; where they diverge, a single trial design risks rejection in one region; where one agency is silent on a point the other regulates, the sponsor must infer obligations. Today this reconciliation is performed manually by regulatory-affairs experts. We introduce cross-jurisdiction regulatory divergence detection: given an FDA requirement and an EMA requirement on the same topic, classify their relationship as AGREE, DIVERGE, or SILENT. SILENT is inherently directional (SILENT_FDA vs. SILENT_EMA); we record direction per pair and report per-direction F1 alongside the collapsed label. We release RegDivergence-101, a 101-pair expert-grounded pilot evaluation benchmark (labels grounded in three peer-reviewed FDA/EMA comparison studies and primary FDA/EMA/ICH guidance text; dual-annotation inter-annotator kappa = 0.85), and systematically characterise a four-method baseline hierarchy: lexical heuristic (0.511 macro-F1, 95% CI [0.411-0.605]), NLI cross-encoder (0.233), obligation-level Graph-RAG (0.663 [0.570-0.747]), and flat LLM judge / Claude Haiku (0.830 [0.747-0.908]). Three directional observations emerge at pilot scale (n = 101): SILENT is semantically detectable but invisible to entailment-only formulations; pair-level obligation graphs improve over lexical methods but trail flat-LLM context (CIs partially overlapping); and corpus-level graph construction is the indicated architectural target for large-scale silent-detection. RegDivergence-101 is a pilot release establishing the task formulation and baseline hierarchy; four unrepresented regulatory domains and an expansion roadmap are described in Section 7.

cs.AI↗

arXiv2026

Risk-based test framework for LLM features in regulated software

Large language models are increasingly embedded in regulated and safety-critical software, including clinical research platforms and healthcare information systems. While these features enable natural language search, summarization, and configuration assistance, they introduce risks such as hallucinations, harmful or out-of-scope advice, privacy and security issues, bias, instability under change, and adversarial misuse. Prior work on machine learning testing and AI assurance offers useful concepts but limited guidance for interactive, product-embedded assistants. This paper proposes a risk-based testing framework for LLM features in regulated software: a six-category risk taxonomy, a layered test strategy mapping risks to concrete tests across guardrail, orchestration, and system layers, and a case study applying the approach to a Knowledgebase assistant in a clinical research platform.

cs.SE↗

arXiv2025

Adhera: A Human-Centered Health Informatics Solution for Reducing Informal Caregiver Burden through Improved Medication Adherence

The growing global population of older adults, combined with ongoing healthcare workforce shortages, has increased reliance on informal caregivers, including family members and friends who provide unpaid support to individuals with chronic illnesses. Among their daily responsibilities, medication management remains one of the most demanding and error-prone tasks. Non-adherence to prescribed regimens not only undermines patient outcomes but also intensifies caregiver stress, anxiety, and fatigue. Although digital health technologies have proliferated to address adherence, most solutions focus exclusively on patients and neglect the informational and emotional needs of caregivers. This paper introduces Adhera, a caregiver-inclusive health informatics system designed to support medication adherence while reducing caregiver burden. Using a mixed-methods research design that included fifteen semi-structured caregiver interviews, sixty-five survey responses, and five pharmacist consultations, this study identified three primary challenges: caregiver stress related to uncertainty about medication intake, fragmented communication with healthcare professionals, and distrust in existing digital tools. Informed by the CeHRes Roadmap 2.0 and the Triple Bottom Line by Design and Culture (TBLD+C) framework, as well as recent co-design studies involving caregivers, Adhera integrates a sensor-equipped smart pill organizer with a mobile companion application that records intake events, sends real-time reminders, and provides caregivers with synchronized adherence data. Preliminary evaluation suggests that Adhera enhances visibility, improves caregiver confidence, and streamlines medication routines. This study contributes to the field of health informatics by demonstrating how human-centered design and collaborative frameworks can align technical innovation with empathy-driven care.

cs.HC↗

arXiv2025

Beyond Chat: a Framework for LLMs as Human-Centered Support Systems

Large language models are moving beyond transactional question answering to act as companions, coaches, mediators, and curators that scaffold human growth, decision-making, and well-being. This paper proposes a role-based framework for human-centered LLM support systems, compares real deployments across domains, and identifies cross-cutting design principles: transparency, personalization, guardrails, memory with privacy, and a balance of empathy and reliability. It outlines evaluation metrics that extend beyond accuracy to trust, engagement, and longitudinal outcomes. It also analyzes risks including over-reliance, hallucination, bias, privacy exposure, and unequal access, and proposes future directions spanning unified evaluation, hybrid human-AI models, memory architectures, cross-domain benchmarking, and governance. The goal is to support responsible integration of LLMs in sensitive settings where people need accompaniment and guidance, not only answers.

cs.HC↗